ClearSpeed has lately made a name with its co-processors. Simply a chip working side-by-side with the processor and make it possible to execute even more floating point calculations. Its little chip, with a total of 96 cores (!) working at 250MHz can actually perform more floating point calculations than a quad-core Opteron can; 25 vs 20 gigaFLOPS and that at only a fraction of the power consumption. ClearSpeed has solutions ready that has been displayed with IBM’s dual-Opteron servers, that then was displayed with two dual-chip solutions which offered a total floating point performance of 100 gigaFLOPS. Exactly what AMD is considering here we don’t know, but we expect more information to arrive soon and we have to say that we are quite curious about what the two might be collaborating here. Source: TechReport
Subscribe
Please login to comment
0 Comments
äldsta